Subclonal Copy Number and Tumor Evolution Usage Guide

Overview

A tumor is a mixture of cell populations. When a copy-number change is present in only some cancer cells, bulk sequencing averages it into a non-integer state, and a long non-integer segment is a subclonal copy-number alteration, not noise. Resolving subclonal CN reveals the tumor's clonal architecture; calling whole-genome doubling explicitly is a prerequisite because it rescales every copy number. This skill covers Battenberg (phased clonal + subclonal CN), TITAN (HMM mixture of cell populations), whole-genome-doubling detection and timing, and MEDICC2 copy-number phylogenies.

Prerequisites

R -e "remotes::install_github('Wedge-lab/battenberg')"   # plus 1000G impute reference
R -e "BiocManager::install('TitanCNA')"
conda install -c bioconda -c conda-forge medicc2         # pip alone misses the OpenFST dependency

Inputs: tumour and matched-normal WGS (or WES for TITAN); allele-specific data (logR and BAF at heterozygous SNPs); a 1000 Genomes phasing/impute reference for Battenberg. For evolution analysis, multi-region or multi-sample data.

Quick Start

Tell the AI agent what to do: - "Call subclonal copy number for this tumour-normal WGS pair with Battenberg" - "Run TITAN and estimate the cellular prevalence of clonal clusters" - "Determine whether this tumour has undergone whole-genome doubling" - "Time whole-genome doubling relative to point mutations" - "Build a copy-number phylogeny across my multi-region samples with MEDICC2"

Example Prompts

Subclonal calling

"Run Battenberg on this tumour-normal WGS pair and report which segments are subclonal, with their major/minor states and cell fractions."

"Run TITAN, sweep the number of clonal clusters from one to five, and select the cluster count by model fit."

Whole-genome doubling

"Decide whether this tumour is whole-genome-doubled using absolute allele-specific copy number, and explain why a depth-only profile cannot answer this."

"Time whole-genome doubling for this tumour using mutation copy number."

Evolution and reconciliation

"Build a whole-genome-doubling-aware copy-number phylogeny across my five regions with MEDICC2."

"Battenberg and TITAN disagree on whether a segment is subclonal. Walk through reconciling them."

What the Agent Will Do

Confirm depth and purity are adequate for subclonal resolution
Establish whole-genome-doubling status from absolute allele-specific copy number
Run Battenberg or TITAN to fit clonal and subclonal copy-number states
Report cancer cell fractions; flag mirrored subclonal allelic imbalance
For evolution questions, build a copy-number phylogeny with MEDICC2
Require concordant evidence before declaring a subclone; flag single-segment claims

Tips

A long non-integer segment is a subclonal CNA, not noise; use Battenberg or TITAN, not a clonal-only caller.
Call whole-genome doubling explicitly before interpreting any copy number; missing it halves every copy number and mis-times every mutation.
WGD calling needs absolute allele-specific copy number; depth alone cannot distinguish a doubled genome from a non-doubled one.
Battenberg resolves subclones to ~3% of cells only at adequate WGS depth and purity; treat low-depth subclonal calls as exploratory.
Mirrored subclonal allelic imbalance can look balanced in bulk BAF; phasing or single-cell data is needed to detect it.
A single subclonal segment is a hypothesis; require multiple concordant segments at a consistent cell fraction before declaring a subclone.
Single-region sampling misses spatial subclones; use multi-region data for evolution.

copy-number/allele-specific-copy-number - Clonal allele-specific CN, purity, ploidy
copy-number/copy-ratio-segmentation - Segmentation feeding subclonal callers
copy-number/hrd-scoring - Whole-genome-doubling correction for the LST scar
copy-number/recurrent-cnv - Copy-number signatures including WGD and chromothripsis
copy-number/cnv-visualization - Visualizing subclonal segments and BAF
variant-calling/vcf-basics - SNV calls for cancer cell fraction and WGD timing